My Chat with Jimmy Chen, Founder & CEO of Propel
Bringing software to public interest problems
I have long admired what Propel does in helping millions of Americans on food stamps. I got a chance to speak with Jimmy, CEO of Propel. He has been at it for eight years now. I enjoyed learning about his time-tested and clear-minded perspective on everything from working with the government and his journey in figuring out the product to his philosophy on business models in servicing the low-income population and balancing venture capital-fueled growth with social good. His patience, passion, and persistence in working on hard impactful public interest problems come through in this conversation.
This is my third conversation in exploring different ways startups engage with the public sector as partners to amplify the intended impact of the policies and programs the politicians and regulators have put in place. It is no secret that the public sector is not great at thinking through and designing the user experience. Jimmy shares his thoughts on governments having that blindspot and how that opens up an opportunity for mission-aligned startups like Propel.
Sar: I want to start by talking about the big picture of the relationship between the government and tech companies. It's a hot topic these days. It's pretty controversial, especially when you look at it from the lens of national security, defense, and law enforcement in the broader tech, media, and political zeitgeist. What is your big-picture view on the relationship between tech and government? And how does that tie into what you guys do at Propel?
Jimmy: I would think about the problems you are trying to solve. That is probably an oversimplification, but I characterize the problems in three categories. You've got the private sector, like capitalist problems; how do you get a cab or find a place to stay where the private companies are the ones to solve those problems? Maybe with regulations, but private companies fundamentally are the solution to those.
The second category is problems that are considered the domain of the government. These are not private sector problems. Some industries like the defense you mentioned are in the other category of selling services to the government because you’re helping the government fulfill its obligation of providing public schools, building roads, having missiles, etc.
And the third category is the things in the middle that I would characterize as public interest problems. A public interest problem that we work on at Propel is what are the services available to help low-income families make it through every month? You could argue that it's the domain of the government. You could also argue that the private sector should fix these problems. The reality is somewhere in the middle.
Many tech companies facing government tensions fall in the first category of working on a private sector solution. You see government as a tax, either literally a tax or qualitatively taxed because they're a regulatory compliance problem for you. And they're telling you that you can't do things or that you can't do things the way you want to.
I'm most interested in the problems in the middle, the public interest problems, where the public has a stake in arriving at a good solution. Still, it's not clear if the government should be the one to provide that solution. Companies that operate in this space can't ignore the government, and the government is the financier of some of the efforts in the space. And in some cases, you're competing with the government. As the consumer, you choose one or the other. We build software that helps people use their food stamp benefits, also known as SNAP. And we share the same goals as the administrators and the policymakers who fund and administer that program. Everyone wants this hundred billion-dollar program to work better. Everyone wants low-income families to be able to buy groceries in a respectful and modern way. Everyone wants those dollars to get stretched as far as possible to make as much impact. And when you're talking about the same goals, it's a question of tactics; then it's about who's got the right tactics to make that happen.
It's tough for the government to do user testing, be data-driven, and go through a software interaction cycle. And those are the things private companies can do well. We hope to contribute to these same goals by bringing to bear the tools of the private sector and the muscle of tech companies in building tools that people want.
Sar: It's easy to imagine an alternate scenario where you could have taken a different path. You could have said, "We want to be working on the food stamps problem. We want to help low-income people. We want to increase the impact of this program. The government is not good at software. Let me sell software to the government.” It's crazy to think how a small line item in the federal budget powers your entire market. You chose to build a consumer brand and a business powered by the government instead of selling software to the government.
Jimmy: Yeah. There are clear advantages to deciding to sell software to the government. The most obvious benefit is that your business model is figured out for you. A disadvantage of going down that path is the government has to want to solve that. That means the government has to understand that problem, which means the government has to understand user experience and the direct pain points that consumers face. And it's not that the government's out of touch. I don't want to make that trite point.
At Propel, we initially got a lot of our traction in our app because we built a better way for people to check their EBT (Electronic Benefit Transfer) balance. Instead of having to call the 1-800 number, we built a smartphone app. We now do other things on top of that, but that was how we got our start. If you had asked the policymakers in charge of the budgets for administering the food stamp program nationally or at the state level, hey, I've built you this awesome app that improves the EBT balance checking experience, they wouldn't have bought it. That's because it wasn't a problem that they identified. If you did a Ph.D. on the social safety net in the United States, and you wrote recommendations for policymakers about how to improve the experience of being poor in this country, you probably wouldn't have come up with "we should replace this 1-800 number with a free app instead." It's so in the weeds. Generally speaking, governments buy software top-down.
Sar: They run a massive procurement process.
Jimmy: Right, the procurement process. Does the government know that a software product can meet this consumer need? Are they even going to have an RFP for it?
Sar: This reminds me of 2020; when the pandemic kicked in, there was a massive push for PPP loans in the COVID relief bill to help small businesses stay afloat. Congress figured out the policy and the budget but wasn't good at managing the qualification and delivery of the funds. An entire army of fintech companies tried to become the user interface between the SMBs and the PPP program by simplifying the application process. Propel is trying to do this in a different market with another program and audience.
Going back to your point about the government not recognizing user experience as a problem, I'm curious about how you got started. You still needed their buy-in because you were a modern layer on top of the primary government service.
Jimmy: Totally. Yeah. I would just say there's a difference between getting people in the government to see the value of user experience and getting people in the government to put out a $50 million RFP to procure software. We've succeeded at the former, but not the latter. We're not asking them to build something. We're asking them to be our partners, give us some air cover as we build, provide us with some advice about how to run it, and agree that what we're doing is within the rules and boundaries of how the program works. We are not requesting them to put this in a budget. We're not requesting them to spin a new division to help us.
One of the challenges in the government technology domain is the activation energy for things is extremely high, whether that's looking at contractors who can't be dislodged from their seats because they've been doing it for 20 years or just the process for how things work. Changes in the government come from policy, and that's not a bottom-up way to think about change. This model of private companies that can improve the user experience of government services and solve public interest problems can come in because they can be more bottom-up, more responsive to consumer needs, and take a more flexible approach to those needs.
Sar: Are there no technical dependencies on the program? It's still a government service. What did it look like mechanically? You're not just building the front end; you’re talking with the government systems.
Jimmy: When I started Propel, we began with something slightly different. We helped people enroll in food stamps. In New York City, you couldn't register for food stamps on your phone in 2014. There was a website that only worked on desktop browsers because it used some kind of flash plugin that would crash on mobile browsers at the time.
Sar: Oh wow, this wasn't too long ago, which is sad!
Jimmy: Right, this isn't ancient history. The first thing I did was I went to the city and asked them to fix it. They said no, they were not going to fix it, or it wasn't a big priority. We tried to hack our way around this and find a way to help people enroll in food stamps on their phones. New York City accepted enrollment for the food stamp program via fax. And there was a 10-page PDF you must fax to the city to apply. So our first ever product was a mobile-friendly website with a responsive front end that would ask the same questions. You'd have to fill out that PDF at the end. We'd have you sign with your finger, and we would save it as a PNG. We overlaid it on top of the PDF form and then faxed it to the city. The first 20 times, I manually printed out and walked over to a FedEx and faxed them into the city. We eventually built programmatic faxing through an API.
We stopped doing that at the end of 2015 because we ran into many operational challenges around faxes. Cities told us to stop doing it. We never figured out a business model for it. When you don't sell to a government, you're left to figure out and invent a business model from scratch. I think it informed some of our thoughts around if we don't have a direct relationship with the government, how might we improve a government program? And are there ways for us to build technology on top of existing interfaces to make these things work?
Sar: You were trying your best to interface with the service without interfacing with the people running the service.
Jimmy: Yeah. We weren't trying to break any rules. There already was an accepted path for how to get this data into the government. It's a fax machine, but it's still a path. We tried to build on that established path until we got shut down.
We approached the EBT balance-checking product similarly in the company's early days. We found out that most states have a website that allows people to log in with their personal information to see their balance and transaction history. We facilitated access to those websites in much the same way Plaid and others in the aggregator space have facilitated access to financial services data. We wondered why isn't anyone doing this for EBT cards and government benefits. So that's how we got our start.
Sar: Were you scraping all these state-level websites?
Jimmy: Depends on the state. Every state runs its own EBT card and its SNAP program. We had to explore how to do this across the country in all 50 states. No state provides perfect APIs. We took on a big operational challenge.
We believe that consumer access to data ought to extend to government benefits because government benefits are a proxy for money. It is patterned after how payments in the private sector are supposed to work. If we believe that consumers have the right to their financial data when it comes to banks, we think consumers should have the right to their financial data regarding government benefits. We've operated with the same philosophies and protections you might see from a major financial services aggregator.
Sar: What was evolution like? You figured out you want to focus on enrolling more people in the program. And then you figured out how fax machines were the data input mechanism. You got shut down. And then what happened? How did you get from there to where you are today?
Jimmy: We never figured out how to monetize signing more people to the program. We tried selling to the government, but it didn't work. We tried selling software to nonprofits that do face-to-face social services but don't generally have software budgets. We tried selling software to grocery stores with the idea that grocery stores want more people to sign up for food stamp benefits because they'll spend more at the store. That also didn't work.
Sar: The AOVs are probably not high enough for them to justify this as a marketing expense.
Jimmy: Yeah. So, that didn't work. By trying to sell it to grocery stores, my co-founders and I spent a ton of time in grocery stores, talking to people using their food stamp benefits. One of the stats we found was that 85% of Americans who qualify for food stamps were already signed up. When you go to grocery stores in low-income neighborhoods, in many cases, 40% to 60% of those shoppers are there to spend their food stamps. We started learning about the experience of spending your food stamp benefits.
About 20 million households have EBT cards, about $120 billion in transaction volume each year. It is a major financial instrument in the United States. We went to a lot of grocery stores where the majority of the revenue from the store came from food stamps. People would tell us they had to call the 1-800 number on the back of their card to check their balance. We heard about people who would buy a banana when they went to the store as the first purchase just to know their balance. The bottom of the receipt would have the balance on the card. We talked about banana balance as one of the concepts behind the app. It just seemed like an eminently solvable problem. We built a basic app that shows your balance and transaction history, so you don't have to call that phone number or buy a banana first.
Sar: This required you to integrate with government systems; this is where the actual work begins.
Jimmy: Yeah. We had to navigate the plumbing and learn how these EBT portals work to build the first version of helping consumers access their information through those portals.
Sar: Are these prepaid debit cards funded by the government? How does it work? Do you handle the flow of funds?
Jimmy: We don't handle the flow of funds for EBT. We ended up building a different product that does handle the flow of funds for other parts of people's financial lives. EBT cards are prepaid debit cards, but they sit on their own somewhat weird rails because they are specific to how you can spend at grocery stores authorized by the federal government and only on certain SKUs
Sar: You mentioned the business model multiple times. I'm curious how you think about business models philosophically and tactically. Many consumer fintech companies target a similar audience and face problems while trying to make money. They sometimes unintentionally start abusing the very people they were trying to help. Many early wage access startups have gotten into legal troubles, for instance. How did you think through the business model? You don't want to make money off the very low-income Americans you're trying to help in the first place. Their entire problem is they don't have the money. Also, we have seen most of the PFM apps struggle. They inevitably hit the ceiling and have spiky early adoption curves. Many high-income people pick them up and lose interest over time, and the problem with low-income people is they don't have enough money to warrant apps to manage the funds.
Jimmy: I agree with your take on PFMs. I do see it a little differently. You are right PFMs have problems with retention and also have high CAC. Those two combined are not going to work. You need incredibly high revenue to overcome that. At Propel, the difference for us is that our CAC is extremely low. It's easy to acquire new customers because we solve actual pain points they face, and we're not in a very competitive industry. There are just not a lot of players in our market. Our retention's extremely high because there are very few other options, and we put a lot of care into our products.
Sar: And it's not a commodity feature across consumer fintech.
Jimmy: Yeah. What we do is differentiated, and that creates value. You asked about how we think about our revenue model. The core principle that I try to follow here is that there are ways to do good things for people and there are ways to make money, and there are some things in the middle, and depending on the industry you're in, there may be no way to make money doing good things for people, right? We are in an industry where we think there is something in the middle, we're going to figure out what those things are, and that's going to be our business model.
We exclusively want to do things that help our users financially and make us money. It's easier to do just one or just the other. You could be a charity or a company that gets regulated for doing bad things. Ideally, we have a business model whereby making revenue creates a positive impact for our users, and by creating a positive impact for our users, we are driving revenue.
One of the challenges that we have accepted at Propel, I've been doing this for eight years, and we're nowhere near where I think we can be as a company, is that we are going to need to combine multiple business ideas to make this work. I would love it If we could make a billion dollars in revenue from a single model. The reality is that the market we're in doesn't support that. We'll need to invent several ways to make money to meet that intersection of both good for users and company revenue.
There are two ways we make money right now. One is lead generation, which helps save money for our users and make money from third parties. That is the classic PFM business model. The difference is that we are more targeted because all our users are on food stamps. So our users are often getting exclusive discounts through us. They're finding jobs through us. They're accessing other government benefits that give them free things through us. That's the alignment of impact and the business model we've tried to achieve.
The other way we make revenue is that we're the program manager for the debit cards. The debit card is intended to supplement the government benefits that someone gets for many people who use our products; they get an EBT card with their food stamps. They may also receive another government cash payment, whether that's a disability payment or an unemployment payment, or a tax refund. We've heard from our users about the pain points they face in receiving those payments as someone who's very low income, who maybe doesn't get a consistent paycheck from an employer, but is looking for a deposit product that meets their specific needs. We earn interchange fees primarily off of that.
Sar: So outside of food stamps, where the government gives people a dedicated debit, your customers might get a whole host of benefits from the government that is not tied to a specific debit card. And that's your opportunity to become the generalized debit card for running their day-to-day lives.
Jimmy: Yeah. The EBT balance can be thought of as an acquisition hook. We have more than 5 million monthly active users now using this product. More than 4 million are our weekly active users. There's a lot of engagement.
The banking piece was a big unlock for us. These are customers who are getting government direct deposit payments in many cases or who are navigating these different safety net programs to get payments. We have felt that even though fintech for the underserved has grown as a sector, it still hasn't grown to be able to meet the needs of low-income families.
Sar: Let’s switch gears. You guys do these engaging user surveys and publish monthly reports. Can you talk about that?
Jimmy: We started doing that in the early days of the pandemic. In March-April 2020, when businesses were shutting down, there were many questions about what would happen. We heard obvious answers from our user base about what was happening. We saw people lose jobs at a high rate. We saw people struggle to go and buy masks and hand sanitizer. And we just saw an opportunity to tell the world what we're hearing from people. We can share what is going on month-to-month for low-income people struggling financially. We still do that on an ongoing basis. We publish these monthly surveys.
What are the trends? There are often these national narratives about unemployment and how different things are happening. We want to tell a more focused story on what's happening for low-income families. Are their lives generally getting better or generally getting worse month to month? There's much talking about how we can help the underserved but not much about describing what earning $10K a year is like.
Sar: Yeah, that's great. I find them educational. If a bunch of your friends told you they wanted to do what you did, but for some other domains, what would you recommend? I strongly believe that designing great software maximizes the social impact of government programs already on the books. As an immigrant, I would pick the immigration process. It's a total nightmare. What are some whitespaces?
Jimmy: Have you heard of Boundless? You should look them up. They are the closest thing to Propel for immigration. One of the challenges with immigration is that it's deeply legal, right? In the company's early days, I was convinced we wanted to build software to help improve the user experience of a safety net program. But we weren't sure which one it was going to be. We picked the food stamp program because of how big the program was and the qualitative arguments around Maslow's hierarchy of needs. You have to start with food before you can do the other stuff. There are 36 government benefit programs. Each of those is its standalone business opportunity, and the company looks very different if you built Propel for Medicaid, Propel for housing, or Propel for utility assistance. Those are all big industries where the user experience that consumers, especially underprivileged consumers, tends to be not very good. These are massive programs that have a significant impact on people. And if I can help someone save a few minutes or improve the user experience of signing up for or getting this thing, it'll be positive for those who use it.
Sar: Are there things you are glad you didn’t know in the early days because that might have stopped you from working on this?
Jimmy: Oh, absolutely! In the early days, Propel was very much driven by naivete. It just seemed crazy that government websites didn’t work better and that you couldn’t check your EBT balance on a smartphone app. It seemed like an obvious area of consumer need and an opportunity to build something 10x better than the status quo by applying consumer software best practices. If I had been a career government administrator, I’m not sure I would’ve seen it the same way. I would’ve probably better understood the systems at play, the bureaucracy, how slow it is to change, and the complexity of the underlying machinery that runs a program as big as SNAP. I got lucky by picking the right set of problems to be naive about. And ultimately, we were correct that people prefer using well-built software to manage their government benefits.
Sar: Is there a recurring topic that keeps coming up internally and causes a heated debate?
Jimmy: One recurring debate is around paternalism in serving our user base. Since we are the most popular digital channel to reach low-income Americans, we often debate what our role ought to be in shaping the content they see. Is it our role to teach people to budget and eat vegetables? Or should we be laissez-faire and be willing to show ads for payday lenders and cigarettes if they pay us?
The reality is usually somewhere in the middle. We risk losing our users’ trust if the platform comes across as preachy. On the other hand, we feel obligated not to lend our megaphone to products that are pretty clearly harmful. Where that exact line gets drawn is always up for debate - and we encourage everyone at Propel, no matter their job function or seniority, to share their views.
Sar: Yeah, I agree the reality is in the middle, and taking a purist stance on either side has its own set of arguably greater problems. Here's my last, maybe touchy, question. How do you balance staying true to the demographic you’re serving with delivering on the venture growth story you have signed up for?
Jimmy: I feel strongly about wanting to be held to the same expectations as our peers in various venture portfolios. That's how you build something that can last if you hold yourself to high expectations around those things. We have investors and a board that I'm pretty proud of. I think it's a question of what timeframe we are talking about. We're trying to build a big, generational organization that will significantly impact poverty in the United States. And that's not going to be something you do overnight. It's also not something you do by cutting corners on trust, right? We are trying to build an institution similar to USAA, a Fortune 100 company. Instead of veterans, we want to focus on low-income families and create something that's that deep, trusted, and institutional. We have to take this long-term view; we will need to build multiple products they will need to make on top of each other. We will need to understand the demographic we’re serving deeply and why they have a unique set of needs. We'll need to build a brand over time, but that's what our investors are excited about. That is the venture growth story for us.
Thanks for reading SarHaribhakti's Newsletter! Subscribe for free to receive new posts and support my work.